JPH07220394A - Audio editing method - Google Patents

Abstract

(57) [Abstract] [Purpose] An object of the present invention is to provide an audio editing method that prevents sudden changes in the sound at the joint when two data are connected. [Structure] From the data encoded in step SP1,
The last n samples of data are taken out and decoded.
A fade-out process is performed on the data X [k] decrypted in step SP2. The first n samples of data are extracted from the data encoded in step SP3 and decoded. A fade-in process is performed on the data Y [k] decrypted in step SP4. In step SP5, the data F0 [k] and the data FI [k] are merged. The data Z [k] merged in step SP6 is encoded. Original data in step SP7,
The cross-faded data E [k] is rearranged into one data.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an audio editing method for smoothly connecting two audio signals.

[0002]

2. Description of the Related Art Conventionally, in audio editing technology,
There has been a technique of connecting two pieces of digital audio data into one piece of digital audio data. In this case, if the two pieces of data are simply connected back and forth, the sound may change abruptly at the joint of the data during reproduction, and it may be difficult to hear.

Therefore, in such a case, a technique called crossfade was used. Crossfade is to gradually decrease the gain of the last part of the previous data and gradually increase the gain of the first part of the latter data.
It is a technology that adds both.

In the audio editing technique,
Miniaturization of equipment by adoption of digital system, multi-channel digital signal and digital signal processing,
In order to realize multimedia using a plurality of media such as tapes and magneto-optical disks, and further to reduce the cost, audio high-efficiency encoding technology that reduces the number of bits to be used without sacrificing sound quality is widely used.

For high-efficiency audio coding, predictive coding for predicting the next sampled value from the past sampled value and coding the difference between the true sampled value and the predicted value, bias of the frequency component, and further Includes transform coding and band division coding that also utilize auditory characteristics.

In such a conventional audio editing technique, there is an editing system for recording two encoded digital audio data as a continuous data as they are, but each encoded data is recorded. There was no editing method of decoding, crossfading the joint portion, re-encoding, and recording.

Further, two pieces of digital audio data coded by the band division coding method are returned to band-divided data, cross-fading is applied to a joint portion of each band, and then re-encoded. There was no edit method of recording.

[0008]

In such a conventional audio editing technique, an editing method of decoding encoded data, cross-fading the joint portion, re-encoding, and recording is provided. Therefore, when the two encoded data are directly combined into one piece of data, the sound abruptly changes at the joint portion, which is inconvenient to hear.

Further, in the conventional audio editing technique, this is equivalent to applying a common crossfade to each band, and therefore, when connecting digital audio data encoded by the band division encoding method, Converting the converted data into band-divided data first, further band-dividing it, decoding it completely, applying a crossfade, re-dividing the band, and then encoding and recording it is not necessary. There was an inconvenience that the processing increased.

The present invention has been made in view of the above circumstances, and provides an audio editing method in which, when two pieces of data are combined into one piece of data, the sound is prevented from changing rapidly at the joint. To aim.

[0011]

The audio editing method of the present invention, as shown in FIGS. 1 and 2, shows the last sample of the preceding data and the data of the following for two encoded digital audio data. Decode the first sample and, respectively, and gradually lower the gain of the last sample of the previous data, while gradually increasing the gain of the first sample of the latter data, and add the joints to create a continuous sequence. The data is re-encoded and then recorded.

In the audio editing method of the present invention, as shown in FIGS. 3 to 4, each frequency band-divided by a band-division band-pass filter and quantized by a quantizer at Nyquist intervals to become a low-frequency signal. For two digital audio data coded by a band division coding method in which a band signal is coded by a coder, the last few samples of the previous data and the first few samples of the latter data are inversely quantized. Conversion to band-split data, gradually lowering the gain of the last few samples of the previous data, while gradually increasing the gain of the first few samples of the latter data, After the parts are added together to form one continuous data, it is re-encoded and recorded.

In the audio editing method of the present invention, as shown in FIGS. 3 to 4, the band division bandpass filter 21 is used.
0 and 211 are for dividing the band into at least two or more.

[0014]

According to the present invention, the last few samples of the preceding data and the first few samples of the following data are respectively decoded from two encoded digital audio data, and the joint portion is cross-faded. By re-encoding after recording and recording, it is possible to soften the abrupt sound change at the joint.

According to the present invention, the last few samples of the previous data and the first few samples of the latter data are band-divided for the two digital audio data coded by the band-division coding method. Data, cross-fading the joints for each band, and then re-encoding and recording to realize cross-fading without performing band division and band combining processing. You can soften sudden changes in sound.

Further, according to the present invention, since the band division bandpass filters 210 and 211 divide the band into at least two or more divisions, it is possible to encode the signal which has become the low frequency signal by the band division encoding method. You can

[0017]

1 shows a schematic block diagram of an embodiment of an audio editing method according to the present invention. In FIG. 1, encoded digital audio data is stored in the memories 100 and 101, and the memory 10 is added after the digital audio data stored in the memory 100.
The digital audio data stored in 1 is connected to form a continuous data.

Data for n samples is used for the crossfade. That is, the last n samples of the encoded data are retrieved from the memory 100 and provided to the decoder 102. From memory 101,
The data for the first n samples of the encoded data is extracted and supplied to the decoder 103.

The decoders 102 and 103 decode the supplied data, and the decoded n pieces of data are supplied to the fade-out processing unit 104 and the fade-in processing unit 105, respectively. Fade-out processing unit 104
Fade-in processing unit 1 and n data processed in
The n pieces of data processed in 05 are stored in the merge processing unit 106.
Is supplied to.

In the merge processing unit 106, 2n pieces of data are merged into n pieces of data and supplied to the encoder 107. The encoder 107 re-encodes the n pieces of data and supplies the data to the data combiner 108. In the data combining unit 108, the last n pieces of data in the memory 100 are removed, the n pieces of data supplied from the encoder 107, and the first n pieces of data in the memory 101 are removed in order. As a result, a series of data 109 is output.

An embodiment of the audio editing method according to the present invention operates as follows. First, when two pieces of encoded digital audio data are connected, if the two pieces of encoded data are directly recorded as one continuous data, the sound abruptly changes before and after the joint, which makes it difficult to hear. . Further, if crossfading is directly applied to two pieces of encoded data, a series of meaningless numerical values will be formed when the data is decoded.

Therefore, according to one embodiment of the audio editing method of the present invention, the last few samples of the preceding data and the first few samples of the following data are respectively decoded with respect to the two pieces of encoded data, and the joint is formed. The cross-fading is applied to the portion of (1) and then re-encoded and recorded.

In this way, when the two data are connected, the preceding sound gradually decreases and the subsequent sound gradually increases, so that abrupt changes in the sound can be softened before and after the joint. can do.

FIG. 2 shows a flow chart for explaining a concrete operation of the embodiment of the audio editing method according to the present invention. In FIG. 2, first, in step SP1, the encoded digital audio data C [i] (0 <i <l-1, i = 0, i = l-1) is encoded.
From the last n samples of data C [j] (ln- <
j <l-1, j = l-n, j = l-1) is taken out and decoded (X [k], 0 <k <n-1, k = 0, k = n).
-1).

At step SP2, the decoded data X [k] is faded out by the following equation (FO [k], 0 <k <n-1, k = 0, k = n-.
1).

[0026]

## EQU1 ## FO [k] = X [k] * (n-1) / (n +
1)

At step SP3, the encoded digital audio data D [i] (0 <i <m-1,
The data D [j] (0 <j <n-1, j = 0, j = n-1) for the first n samples is extracted from i = 0, i = m-1) and decoded (Y). [K], 0 <k <n-1, k
= 0, k = n-1).

In step SP4, the decoded data Y [k] is subjected to a fade-in process by the following equation (FI [k], 0 <k <n-1, k = 0, k = n. −
1).

[0029]

FI [k] = Y [k] * (K + 1) / (n + 1)

At step SP5, FO [k] (0
<K <n-1, k = 0, k = n-1) and fade-in processed data FI [k] (0 <k <n-1, k =
0, k = n−1) is merged by the following equation (Z
[K], 0 <k <n-1, k = 0, k = n-1).

[0031]

## EQU3 ## Z [k] = FO [k] + FI [k]

In step SP6, the merged data Z [k] (0 <k <n-1, k = 0, k = n-1)
Is encoded (E [k], 0 <k <n-1, k = 0, k
= N-1).

At step SP7, the original data C
[I] (0 <i <l-1, i = 0, i = l-1), D
[I] (0 <i <m-1, i = 0, i = m-1), cross-faded data E [k] (0 <k <n-1, k
= 0, k = n-1) are rearranged as follows to form one data (R [r]], 0 <r <l + m + n-1, r
= 0, r = 1 + m + n-1).

[0034]

## EQU00004 ## R [r] = C [r] (0 <r <l-n-1, r
= 0, r = 1-n-1)

[0035]

## EQU00005 ## R [r] = E [r- (ln)] (ln <r
<L-1, r = l-n, r = l-1)

[0036]

## EQU6 ## R [r] = D [r + n-1] (l <r <l + m
-N-1, r = l, r = l + m-n-1)

FIG. 3 shows a schematic block diagram of another embodiment of the audio editing method according to the present invention. The block diagram shown in FIG. 3 is different from the block diagram shown in FIG. 1 in that two pieces of encoded data are divided into bands by band division band pass filters 210 and 211, and quantizers 212 and 213 are used. Is performed by the inverse quantizers 202 and 203, and inverse quantization is performed by the inverse quantizers 202 and 203.

The process of applying a crossfade to the original signal is equivalent to applying a common crossfade to each band, so two data encoded by the band division encoding method are As described in the prior art, as compared with the case of completely decoding, crossfading, then re-encoding, and recording, a new band combining process and band dividing process are performed in the middle. No additional signal processing is required.

Therefore, according to another embodiment of the audio editing method of the present invention, the last few samples of the previous data and the first few samples of the latter data of the two data coded by the band division coding method are used. Are converted into band-divided data, cross-fading is applied to the joint portion for each band, and then re-encoded and recorded. By connecting the two data in this way, it is not necessary to add new signal processing such as band synthesis processing and band division processing.

FIG. 4 shows a flowchart of another embodiment of the audio editing method according to the present invention. In this case,
In steps SP10 to SP14, the coded band components are not completely decoded but simply returned to the band components, and the above-described processing is performed for each band.

That is, after starting, step SP10
In, the band division is performed on the two encoded data. In step SP11, the data divided into each band is quantized. In step SP12,
After dequantization, a fade-out process is performed in step SP13, and in step SP14, a dequantization is performed and then a fade-in process is performed in step SP15. The subsequent processing from step SP15 to step SP18 is the same as the processing from step SP4 to step SP7 shown in FIG.

In addition, the band division band pass filter 21
Bands 0 and 211 may be divided into at least two or more bands.

[0043]

According to the present invention, the last few samples of the preceding data and the first few samples of the following data are respectively decoded from the two encoded digital audio data, and crossfading is performed at the joint portion. By re-encoding and recording after applying, it is possible to soften the sudden change in sound at the joint.

Further, according to the present invention, for the two digital audio data coded by the band division coding method, the last few samples of the previous data and the first few samples of the latter data are band-divided, respectively. Data, cross-fading the joints for each band, re-encoding, and recording to realize cross-fading without performing band combining and band combining processing. You can soften sudden changes in sound.

Further, according to the present invention, since the band division band pass filter divides the band into at least two or more divisions, it is possible to encode the signal which has become the low frequency signal by the band division encoding method.

[Brief description of drawings]

FIG. 1 is a block diagram of an embodiment of an audio editing method of the present invention.

FIG. 2 is a flowchart of an embodiment of the audio editing method of the present invention.

FIG. 3 is a block diagram of another embodiment of the audio editing method of the present invention.

FIG. 4 is a flowchart of another embodiment of the audio editing method of the present invention.

[Explanation of symbols]

 100 Memory 101 Memory 102 Decoder 103 Decoder 104 Fade-out Processing Unit 105 Fade-In Processing Unit 106 Merge Processing Unit 107 Encoder 108 Data Combining Processing Unit 109 Single Data 210 Band Division Band Pass Filter 211 Band Division Band Pass Filter 212 Quantizer 213 Quantizer 200 Memory 201 Memory 202 Inverse Quantizer 203 Inverse Quantizer 204 Fade Out Processing Unit 205 Fade In Processing Unit 206 Merge Processing Unit 207 Encoder 208 Data Combining Processing Unit 209 Series of Data

Claims (3)

[Claims] 1. For two encoded digital audio data, the last sample of previous data and the first sample of subsequent data are respectively decoded, and the gain of the last sample of previous data is gradually reduced. At the same time, the gain of the first sample of the subsequent data is gradually increased, and the joints are added together to form one continuous data, which is then re-encoded and recorded. . 2. A band division coding method in which a signal of each frequency band that has been band-divided by a band-division band-pass filter and quantized by a quantizer at Nyquist intervals to become a low frequency signal is encoded by an encoder. For the two digital audio data encoded by, the last few samples of the previous data and the first few samples of the latter data are dequantized, respectively, and band-divided data is returned. Gradually lower the gain of the last few samples, and gradually increase the gain of the first few samples of the latter data, add the joints of each band into a continuous data, and then re-encode. , An audio editing method characterized by being recorded. 3. The audio editing method according to claim 2, wherein the band division bandpass filter divides the band into at least two or more divisions.